Democracy and Development

Carolina Torreblanca

University of Pennsylvania

Global Development: Intermediate Topics in Politics, Policy, and Data

PSCI 3200 - Spring 2025

Agenda

Democracy and Development

What is this plot showing?

Two explanations

  1. Causal

  2. Correlational

Development causes democracy

Modernization Theory

Democracy Spurs Development

Democracy and Development in Research

The theoretical links btw democracy and development have permeated micro-development research

Resource booms and democracy - Windfalls do not improve democratic quality but might make autocracies sturdier

Corruption and Development - - Horizontal accountability random audits of Brazilian majors reduce reelection of corrupt politicians

Female participation and development - Looks at random gender quotas in India. Having a female local politician improves development outcomes for female

Democratic quality and deforestation - Democratic competition increases deforestation!

All these literatures borrow from modernization theory

Acemoglu et. al. 2008

The criticism: correlation of democracy and development does not mean causation

Alternative: Countries embark on divergent development paths

Empirical implication: Observed association is correlational

Which is the correct DAG?

How to decide?

We need to control for U

Impossible

We can control for everything

Simpson Paradox

Fixed effects

Simple idea: control for “belonging to a group” using group-level Fixed Effects

  • A way to control for every unobserved group characteristic that does not change

  • In practice: give an intercept to each group

Fixed effects

Go from

\[\begin{equation*} Democracy_{it} = \alpha + \beta_1 Democracy_{it} + \epsilon_{it} \end{equation*}\]

To

\[\begin{equation*} Democracy_{it} = \alpha + \beta_1 Democracy_{it} + \mu_1 USA + \mu_2 France + ... \mu_n Zimbabwe + \epsilon_{it} \end{equation*}\]

Fixed effects: Empirical example

pacman::p_load(readxl, tidyverse,  here, estimatr)

d <- read_xls(here("./slides/code/Income-and-Democracy-Data-AER-adjustment.xls"), sheet = 2) %>% 
  filter(is.na(lrgdpch)==F, 
         is.na(polity4)==F) %>% 
  arrange(country, year) %>% 
  group_by(country) %>% 
  mutate(lag_dem = lag(polity4), 
         lag_income = lag(lrgdpch)) %>% 
  ungroup()
# Polity2 Index
summary(d$polity4)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  0.000   0.150   0.700   0.573   0.950   1.000 
# Log real GDP per capita (PWT)
summary(d$lrgdpch)
   Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
  6.094   7.297   8.149   8.167   8.960  10.413

Fixed effects: Empirical example

ggplot(d, aes(y = polity4, x = lag_income)) +
  geom_point() +
  geom_smooth(method = lm, se=F)

Fixed effects: Empirical example

mod <- lm_robust(polity4~lag_dem + lag_income, clusters = country, data = d)
modelsummary::modelsummary(mod, stars = T, coef_omit = "(Intercept)",  gof_omit = "AIC|BIC|Log.Lik|F|R2")
Model 1
lag_dem 0.742***
(0.032)
lag_income 0.063***
(0.010)
Num.Obs. 900
se_type CR2
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001

Add Fixed Effects

Add Fixed Effects

require(estimatr)
mod2 <- lm_robust(polity4~lag_dem + lag_income, fixed_effects = country + year, clusters = country, data = d)

modelsummary::modelsummary(mod2, stars = T, coef_omit = "(Intercept)",  gof_omit = "AIC|BIC|Log.Lik|F|R2")
Model 1
lag_dem 0.431***
(0.054)
lag_income −0.005
(0.033)
Num.Obs. 900
se_type CR2
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001

Summing up

  • Good theoretical reasons to think

    • Democracy is causally related to devel

  • Most extant work focuses on the micro, subnational dynamics

  • But the broad theoretical argument is old!

  • Fixed effects can help control for time invariant unobserved confounders

  • If we think confounders might change in time, for exampe